The pebble- to cobble-sized basalt clasts found in Early Cretaceous (Barremian- to early Aptian-aged) claystone and calcareous claystone below the thick breccia unit in Hole 899B exhibit a range of textures and vesicularity. Some basalt clasts are vesicular, and the vesicles are often filled with secondary white or pink minerals. Some basalt clasts have obvious discolored weathering rinds. Part of a highly altered autobrecciated basalt clast (interval 149-899B-26R-1, 115-118 cm) was analyzed in this study but no thin section was made because this friable rock fragmented badly during transport. The diabase clasts were called microgabbro clasts in the Initial Reports volume (Sawyer, Whitmarsh, Klaus, et al., 1994) because of a lack of obvious plagioclase laths regarded as characteristic of a diabasic texture. However, the diabase clasts have relatively short plagioclase laths and intergranular textures when observed in thin section and should be called diabase clasts rather than microgabbro clasts. Only ten nonvesicular and nonbrecciated basalt and diabase clasts were sampled for analysis and only eight of the ten clasts were thin sectioned. The five basalt clasts that were thin sectioned exhibit three distinctive textures that indicate different cooling histories and suggest derivation from separate sources, whereas the three diabase clasts have similar textures.
Two of the five thin-sectioned basalt clasts (interval 149-899B-27R-1, 19-26 cm, and interval 149-899B-29R-1, 99-102 cm) are porphyries in which 20% to 25% of the clasts consist of a range of large to small, twinned plagioclase phenocrysts with occasional zoning (Fig. 4). A microprobe traverse across one of the phenocrysts in a basalt clast from interval 149-899B-27R-1, 19-26 cm, indicates it has a relatively uniform composition near An80 (Fig. 5) and a low K content (An80.76, Ab19.07, Or0.17). The plagioclase phenocrysts are relatively fresh in the basalt clast from interval 149-899B-27R-1, 19-26 cm, but are more badly altered in the basalt clast from interval 149-899B-29R-1,99-102 cm. What appear to be completely replaced olivine phenocrysts also occur in both of these porphyries. Two other basalt clasts (interval 149-899B-27R-1, 13-15 cm, and interval 149-899B-31R-1, 9-14 cm) have spherulitic to variolitic textures and the Initial Reports volume (Sawyer, Whitmarsh, Klaus, et al., 1994) refers to their variolitic texture. The remaining four clasts, a basalt clast from interval 149-899B-27R-1, 33-40 cm, and three diabase clasts from interval 149-899B-34R-1, 58-62 cm, interval 149-899B-35R-1, 0-7 cm, and interval 149-899B-36W-1, 37-42 cm, all have intergranular textures. Interval 149-899B-27R-1, 33-40 cm, contains a fine grained intergranular textured basalt that is badly altered with silicification and oxidation obscuring much of the original mineralogy. The other three intergranular textured clasts are coarser grained diabases that were called microgabbro clasts in the Initial Reports volume (Sawyer, Whitmarsh, Klaus, et al., 1994). These diabase clasts consist of 50% to 60% short plagioclase laths surrounded largely by fragments of clinopyroxene in various stages of alteration. The alteration material includes acicular masses of actinolite, brown biotite, and locally abundant iron oxide stain. The coarser diabase clasts must be derived from a flow interior, a dike, or a sill. The variable textures of the basalt clasts suggest a variety of sources rather than a single source. By contrast, the diabase clasts appear more uniform suggesting a single source. The source or sources of these clasts will be explored by comparing compositional data with that of typical basalts from known sources.
The major element compositional variations observed in the basalt clasts indicate alteration by seafloor weathering. Several oxides vary systematically with changing LOI (Fig. 6) with most oxides decreasing with increasing LOI in what appears to be dilution caused by increasing LOI. However, MgO increases dramatically with increasing LOI while total iron increases slightly, suggesting addition of these components to the rock from the altering solutions. Also the CaO content decreases much more rapidly with increasing LOI than other oxides, suggesting a loss to the altering solutions. The most altered basalt clast (interval 149-899B-26R-1, 113-118 cm), in which LOI = 13.4%, has a much altered composition, with CaO = 0.76%, A12O3 = 11.24%, and MgO = 28.66%, when calculated dry. This argues strongly for removal of CaO, addition of MgO, and perhaps some removal of A12O3. The other basalt clasts, which have lower LOI values, but similar oxidation values, reveal more typical basalt compositions when calculated dry, except for CaO and MgO which are unusually low and unusually high, respectively, and show a strong negative correlation (Fig. 7). The influence of the added MgO can be observed on an alkali-iron-magnesium (AFM) diagram (Fig. 8), where the basalt clasts plot off the high MgO end of the Skaergaard magma evolution trend in a region where cumulates typically plot. Because MgO and FeOt increase together in these altered clasts, the effect illustrates greater addition of MgO rather than depletion of FeOt.
Experimental studies (Bischoff and Seyfried, 1978; Hajash, 1975; Humphris and Thompson, 1978; Mottl and Holland, 1978; Seyfried and Bischoff, 1979; Hajash and Archer, 1980) have determined that altered basalt gains Na and Mg from seawater and loses Ca, Fe, Mn, and Si to seawater at all experimental conditions. The relative extent to which various elements are exchanged between basalt and seawater depends primarily on temperature and the water/rock ratio (Hajash and Archer, 1980). Consequently, the increase in MgO and loss of CaO in basalt clasts occurs at all temperatures and water/rock ratios and provides no information on the conditions of alteration. However, the gain of K2O by the basalt and diabase clasts indicates low temperature alteration, below 200°C (Seyfried and Bischoff, 1979), in agreement with observations of K2O gains from seafloor weathering (Thompson, 1973, Thompson and Humphris, 1977). Above about 200°C, K2O appears to be lost from basalt to seawater (Bischoff and Dickson, 1975; Mottl and Holland, 1978). The relative change in concentrations of other major elements due to seawater alteration is typically less clear.
Variations in major element composition not obviously related to seafloor weathering may reveal some indication of the petrogenesis of the six basalt clasts. Perhaps the most significant variation is shown by TiO2 (0.64% to 1.84%), calculated dry, which does not correlate with variations in the degree of alteration as estimated from LOI content, LOI = 4.4% to 13.4%, and the oxidation ratio, Fe2O3/FeO = 0.37 to 1.10. The basalt clast from interval 149-899B-27R-1, 19-26 cm, has low TiO2 (0.64%), and low K2O (0.31%), compared to the other clasts which all have TiO2, and K2O, values over 1 %, suggesting a distinct source from the other basalt clasts. The low TiO2 clast is also more depleted in incompatible trace elements such as Ba, Zr, and Sr.
Much of the major element compositional variation in the diabase (microgabbro) clasts is similar to that noted in the basalt clasts and can also be attributed to seafloor weathering. Several of the major element oxides decrease progressively with increasing LOI content (Fig. 9), in the manner observed for the basalt clasts, but with less scatter. The smaller degree of scatter in the diabase clasts relative to the basalt clasts could be because of a smaller sample set, or a smaller range of LOI variation, or it could be interpreted to suggest that the texturally similar diabase clasts started with a more similar major element composition and perhaps from a common source. Although CaO values are very low in the diabase clasts, they do not vary systematically with LOI as in the basalt clasts. Instead K2O and FeOt are found to vary systematically in the diabase clasts. Both MgO and FeOt increase with increasing LOI, with MgO again increasing dramatically, suggesting addition by seafloor weathering. The less altered diabase clasts have unusually high but irregular K2O (1.90% to 5.06%), calculated on an anhydrous basis, and unusually low CaO (1.90% to 2.75%). The addition of K2O to the less altered diabase clasts is similar to, but more extreme than, the addition observed in the basalt clasts. However, the most altered diabase clast (interval 149-899B-33R-1, 25-31 cm) with LOI = 12.4%, and a highly altered composition with MgO = 28.43%, SiO2 = 39.21%, calculated on an anhydrous basis, has a low K2O value of only 0.36%. This clast also has apparent small increases in FeOt, MnO, P2O5, a large increase in Zr and, more significantly, a low abundance of trace elements Ba and Sr which are typically companion travelers with K. Perhaps a second event removed these soluble cations.
Trace elements provide greater resolution for distinguishing between basalts from various provinces than major elements because of their greater variability. Because these basalt and diabase clasts occur in the transition zone between ocean and continent, they can have either a continental or oceanic source. Among the oceanic basalts, the mid-ocean ridge basalts (MORB) can be distinguished from ocean-island basalts (OIB) and island-arc basalts (IAB) by their elemental and isotopic concentrations. The ocean-plateau basalts (OPB) have elemental and isotopic characteristics that overlap with both MORB and OIB, but in most respects are more similar to MORB (Floyd, 1989; Castillo et al., 1991). However, no isotopic information has been obtained for these clasts, and comparisons must be based entirely on major and trace element compositions and compositional patterns. One of the more useful diagrams for identifying the source environment of basalts has been variably called a spider diagram, a spidergram, or an extended Corell-Masuda diagram because it involves several of the REEs. Sometimes the diagram is referred to as a spider diagram and the elemental data plot or plots on the diagram spidergrams. This usage will be followed here.
The sequence of major and trace elements used in spider diagrams reflects the bulk distribution coefficient or incompatibility of the elements in basaltic magma. Varying interpretations have resulted in a variety of elemental sequences (Sun et al., 1979; Pearce, 1982; Thompson et al., 1983; Hofmann et al., 1986; Sun and McDonough, 1989; Pearce and Parkinson, 1993), but the sequence is intended to produce systematic variations in incompatibility that result in smooth spidergram patterns for MORB; and consequently also for OIB. The spider diagram used here follows the elemental sequence of Sun and McDonough (1989), although K was exchanged with Nb and Ta to produce a smoother MORB pattern, and Cs and Pb have been omitted because many rocks lack these data. Sun and McDonough (1989) indicate that the order of elements U, Nb, Ta, and K is somewhat optional by indicating they have essentially equivalent incompatibilities. Figure 10 shows the resulting spidergrams for normal MORB (N-MORB), enriched MORB (E-MORB), OIB, and continental flood basalt (CFB), normalized to primitive mantle also as defined by Sun and McDonough (1989). These spidergrams stand in marked contrast to the irregular pattern produced by typical island-arc tholeiites and island-arc tholeiites-alkali basalt (IAT and IAT-AB) in Figure 11. The values used for typical IAT have been based on values by Sun (1980) and other studies, with the range of compositions expressed by two spidergrams; one for IAT and another for IAT-AB.
A spidergram was derived for CFB, the most abundant type of continental basalt, to allow comparison with the various oceanic basalts. The resulting CFB spidergram represents an average of several flood basalts judged to be relatively free of contamination from continental lithosphere on the basis of published isotopic and trace element analyses (Bailey, 1989; Brannon, 1984; Dosso, 1984; Hooper and Hawkesworth, 1993; Lightfoot et al., 1993; Paces and Bell, 1989; Seifert et al., 1992; Wooden et al., 1993). The typical uncontaminated, or more likely slightly contaminated, CFB has a spidergram pattern rather similar to E-MORB, except for a Ba peak and a P low, with an elemental abundance between typical E-MORB and OIB (Fig. 10). The validity of the Ba peak and P low was checked by plotting individual flows used to derive typical CFB. They were found to have a wide variety of spidergrams that reflect the wide variability between individual CFB provinces. Continental basalts should be expected to be more variable than oceanic basalts because of their long path through highly variable continental lithosphere. Consequently, even if the two deviant elemental abundances on the typical CFB spidergram are real, and many individual flows have no significant peaks for these elements, the large variation among individual CFB flows limits their value as a distinctive test for a so-called typical CFB source. Furthermore, the similarity between spidergrams for E-MORB and the least contaminated CFB, ignoring the two deviant elements, makes distinction between these sources somewhat equivocal.
A total of six spidergrams has been derived for each basalt clast to allow visual comparison of the clast spidergram with primitive mantle (P-Mantle) and the typical basalt from each of the five major basalt provinces (Fig. 12, Fig. 13, Fig. 14, Fig. 15, Fig. 16, Fig. 17). The overall similarity between IAT and IAT-AB allow comparison with either adequate for identification of an island-arc source province. A perfect match between a clast and a particular source is obtained when all of the elements on the clast spidergram have a value of one. A perfect match would be interesting but no better than a relatively close match because the typical basalt for a given province probably does not actually exist. Comparison of clast spidergrams with the various basalt provinces shows that most basalt clasts exhibit a reasonable match with typical E-MORB or CFB, although the clast from interval 149-899B-27R-1, 19-26 cm (Fig. 12) plots between E-MORB and N-MORB. All clasts show anomalously high K peaks, and often high Rb, relative to that predicted for unaltered E-MORB or CFB. Spidergrams for clasts from interval 149-899B-26R-1, 113-118 cm (Fig. 13) and interval 149-899B-29R-1, 99-102 cm (Fig. 14) fall on the typical E-MORB and CFB abundance lines, whereas the clasts from interval 149-899B-27R-1, 13-15 cm (Fig. 15), interval 149-899B-27R-1, 33-40 cm (Fig. 16) and interval 149-899B-31R-1, 9-14 cm (Fig. 17) plot slightly above that line. Consequently, two samples are essentially equivalent to the typical E-MORB or CFB as plotted, whereas the three samples plotting slightly above the line are somewhat more enriched than typical E-MORB or CFB. Sample 149-899B-27R-1, 19-26 cm, is slightly more enriched than the typical N-MORB, crossing the equivalency line at a small angle, but falling below the abundance expected for E-MORB and CFB. All of the basalt clasts have distinctive Nb-Ta peaks relative to Nb-Ta depleted IAT, and consequently IAT-AB, indicating they are not derived from convergent plate boundaries.
Spider diagrams plot elements with varying characteristics and behavior and can be used to assess the relative mobility of the various elements if all elements have not been mobile. The REEs and high field-strength elements (HFSEs) are aligned on the basalt clast spidergrams and are typically regarded as the least mobile elements in aqueous solutions. High field strength elements on these spider diagrams include Hf, Zr, Ta, Nb, and Ti. Thorium can also behave in a similar manner. The remaining elements are mostly large ion lithophile elements (LILEs), which are notoriously mobile in aqueous solutions. LILEs plotted on these diagrams include K, Rb, Ba, and Sr. The LILEs commonly exhibit anomalous abundance peaks or lows on the basalt and diabase clast spidergrams relative to the less mobile REEs and HFSEs, confirming the mobility of LILEs during seafloor weathering. Uranium also commonly falls significantly off the immobile-element defined spidergram lines for these clasts, indicating it has also been mobile. Basalt clasts from interval 149-899B-27R-1, 19-26 cm (Fig. 15) and interval 149-899B-29R-1, 99-102 cm (Fig. 17) display above average amounts of Sr relative to typical E-MORB and CFB basalt province spidergrams, but less than the typical IAT spidergrams, that correlate with small positive Eu peaks contributed by the plagioclase phenocrysts in these clasts. Anomalous peaks for Sr and Eu are to be expected together in plagioclase phenocryst clasts because the ionic radii of these elements are very similar and both concentrate in the plagioclase lattice.
The REEs are regarded to be the least mobile elements during seafloor alteration processes (Ridley et al., 1994) and should best retain their primary igneous concentrations in these weathered basalt and diabase clasts. Despite the variable textures of the basalt clasts, all of the basalt clasts have smooth REE patterns similar to E-MORB, or less likely CFB, although the basalt clast with low TiO2 and K2O (interval 149-899B-27R-1,19-26 cm) has a flat transitional MORB, between E-MORB and N-MORB, REE pattern (Fig. 18). Both of the basalt clasts with plagioclase phenocrysts (interval 149-899B-27R-1, 19-26 cm, and interval 149-899B-29R-1, 99-102 cm) have the expected small positive Eu anomalies, although the anomaly is best developed in the clast from interval 149-899B-27R-1, 19-26 cm (Fig. 18). Despite the differences in textures, the REE patterns for all of the basalt clasts are similar, with a chondrite normalized abundance of the heavy rare earth elements (HREEs) near 10 times chondrite and the light rare earth elements (LREEs) with an abundance between 10 and 40 times chondrite. The similarity and smoothness of all REE patterns for the variably weathered basalt clasts indicates the REEs have not been significantly affected by seafloor weathering and all of the clasts have a transitional to enriched MORB origin or CFB origin. Obviously the spidergram patterns for E-MORB and least contaminated CFB are so similar that distinction between these basalt types is not easily possible with spidergrams. Isotopic data are needed to distinguish between E-MORB and least contaminated CFB.
The two diabase clasts (interval 149-899B-34R-1, 58-62 cm, and interval 149-899B-35R-1, 0-7 cm) analyzed for trace elements have spidergram and REE patterns similar to each other and to the most trace element depleted basalt clast (interval 149-899B-27R-1, 19-26 cm). The spidergrams for the two diabase clasts are similar, crossing the typical N-MORB line at a small angle and paralleling the E-MORB and CFB reference lines at slightly lower abundances (Fig. 19, Fig. 20). Relative to MORB or CFB, and even the basalt clasts, the diabase clasts are more strongly enriched in K, Rb, and Ba, and slightly enriched in Sr. Elements K, Rb, and Ba were probably enriched during seafloor weathering and entered the secondary biotite observed in the diabase clasts. The enrichment of the diabase clasts in Nb-Ta indicates they are not convergent plate boundary Nb-Ta-depleted basalts. The REE patterns for the two diabase clasts are flat with an abundance near 10× chondrite and small positive Eu peaks (Fig. 21) indicating a transitional MORB or CFB origin.